The Agent Skills Directory

[SAFE]: The skill serves as documentation for a legitimate service from a recognized vendor (Redis). All external resources, including the langcache Python package, documentation links to redis.io, and API endpoints, are vendor-owned and present no inherent security risk.
[SAFE]: The skill correctly demonstrates the use of environment variables (API_KEY, HOST, CACHE_ID) for managing sensitive credentials, avoiding hardcoded secrets.
[INDIRECT_PROMPT_INJECTION]: The skill implements a semantic caching layer, which naturally processes untrusted user prompts and LLM responses, creating a standard surface for indirect prompt injection.
Ingestion points: The prompt parameter in lang_cache.search() and both prompt and response in lang_cache.set() (found in SKILL.md and references/langcache-usage.md).
Boundary markers: Absent; the examples show direct interpolation of prompts and responses into the caching functions.
Capability inventory: The skill performs network requests via the SDK and curl to interface with the Redis Cloud API.
Sanitization: Not present; the skill assumes the application handles sanitization before or after the caching layer.

redis-semantic-cache